diff --git a/docs/index.html b/docs/index.html
index 8609ec4e..7a2765df 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -946,10 +946,11 @@ <h1>Deterministic <span class="highlight">mock LLM</span> server for testing</h1
     <section id="features" class="reveal">
       <div class="container">
         <span class="section-label">Why llmock</span>
-        <h2 class="section-title">Everything you need to test AI integrations</h2>
+        <h2 class="section-title">Stop paying for flaky tests</h2>
         <p class="section-desc">
-          Built for E2E test suites where multiple processes — your app, agent workers, framework
-          runtimes — all need to hit the same mock endpoint.
+          Tests that hit real LLM APIs — OpenAI, Gemini, Anthropic — cost money, time out, and
+          produce non-deterministic results. llmock replaces those calls with immediate,
+          deterministic responses from a real HTTP server any process on the machine can reach.
         </p>
 
         <div class="features-grid">